LMS SITE DROWSINESS PROCTORING

INTRODUCTION

In e-learning environments,  students and teachers use the Internet on a regular basis in order  to follow/receive lectures, ask/answer questions and send/receive assessments. However, e- learning universities rely on an examination process in which students hold a face to face exam in a physical place determined by the university under supervised conditions. Such conditions ensure the correctness of the exam, a difficult task to achieve in a virtual exam model. Face to face exams allow us to check a student’s identity and ensure exam authoring using traditional means (checking an identity card and ensuring no one helps the student during the exam).

In data assortment (for the event of the causal guide), the meeting is utilized as an instrument, a causal guide is formed for each one among the interviewees, and a worldwide causal guide is so created with the conventional evaluation of the specialists, which allows its examination through the FCMappers instrument, that demonstrates the foremost persuasive variables, the parts that get the foremost impact from the rest, and also the most pertinent parts within the framework. Within the related area, electronic deputation instruments in instructive organizations are contextualized through AN abstract survey that includes the circumstance of oversight in distant educating, e-delegating within the instructive framework is contextualized, and also the most compelling. Inspirational parts in tolerating the use of the latest innovative instruments are point by point. we tend to proceed with the approach utilised during this examination moreover, the examination of the outcomes and shut by career attention to the inspirational elements deciding the usage of e-administering in web instructing

Ensure student identity and authoring in a virtual or distance exam has been pointed out as a hard problem in the literature with a difficult solution. Then, e-learning institutions still need face to face exams. However, face to face exams represent an important effort for e-learning institutions. Typically, e-learning universities do not have enough physical facilities for all students so they have to rent buildings in order to allow students to hold their exams. Further- more, exam management becomes more complex since such external examination centers must be provided with all management mechanisms to ensure that students will be able to perform their exam in a desired location and later on, all exam answers will be properly

collected and sent to the teachers that have to correct them. For all those reasons, improving exam man- agement systems has clear advantages for distance learning institutions. Intrinsically, exam management needs to achieve a good security level, since the correctness of this process ensures somehow the quality of the university. For that reason,designing an electronic management system for exams should take special care of security.

Artificial intelligence and Machine Learning have gained increasing popularity in recent years due to their ability to handle tasks that would otherwise take too much computational power, and due to their versatility, the wide range of problems they have been shown to solve. The world is familiar with what E-Learning is. M-Learning has enhanced e-learning by mak- ing the process “learner centered”. With the rise of smartphones and tablets, several million pages of educational material can be shared online. However, the exam part of the curriculum is not very well adapted to this new technology. Executing exam security in an open environment where a student has his/her mobile phone/ laptop connected to the internet where students can easily exchange in- formation can be the most challenging task. This project aims at identifying various vulnerabilities that can violate online exam security and try to resolve all problems and create a more secure on- line exam system. Technology Used: The front end of software will be designed using HTML, CSS,javascript, bootstrap. Machine learning algorithms can be used for face detection and during proctoring. We are using the OpenCV library. MySQL will be used for creating Database.

PROPOSED SYSTEM:-

In this project, we propose to build a system that searches for any person in the image/video who is cheating during examination.System should generate the proper output for every user. We train the model with the dataset using the YOLO algorithm. In this dataset, there are 8000 random images. The model will train on these images and generate the frames for images. When a user logged in proctoring then the camera initialized and started recording. After initializing, the system will throw random and simple questions, which the user has to solve. After that, the examination system captures random 5-6 images and starts working on it. if the system finds any false face or user cheating during examination then the system will throw an alert message. The system also checks multiple login. If the user found that the same username has multiple IP addresses then the system sends an alert message that the user has found multiple logins.

MODULES:-

• Registration: Students who register in a portal for the first time submit their personal data, their ID card and their photo, which is stored in the database and verified using their photo before the exam.
  
  
• Face recognition: A webcam is installed in the a student’s computer or front camera, when the student takes a test on a face recognition recognizes the student  and if the face matches the stored facial image, the student is verified and allowed to take the exam.During the exam, the student’s image is continuously captured and if the face does not match the stored image, their record is saved in the database. Multiple face detection: If there is more than one person  in the picture, this is also recorded in the database.
  
  
• Head Position Detection: For MCQ-based exams that do not require pencil and paper, students’ head position is analyzed and if it appears that a student is looking at the other side of the screen, your dataset is also analyzed be saved.

METHODOLOGY FOR PROCTORING:

1. Front end : The design,layout of the software and how it will interact with the user by using HTML,CSS and JAVASCRIPT.
2. Back end : Back-end will be designed using python Flask server. After creating the front end back end will be developed which contains the mechanism of the application. This will communicate with the databases as well as with the front end languages.
3. Features : To make online proctoring more secure, candidates have to do face verification during login and OpenCV will be used to monitor the moment of the candidate and to These features will be added in back end development and then will be used for proctoring purposes.

This project’s aim is to create Windows application software that can be used to conduct and manage online exams and have better security. The software will reduce the chance of student malicious activities and creates a secure exam environment. This software will have different levels of candidate authentication like Username and Password and face recognition. Teacher will have a separate login to this software which he/she can use to create classes and exams. The no. of students in the class, schedule of exam, exam time and admission of

students the course will remain in hands of the respective teacher.. Students should undergo face recognition to login on software successfully. The teacher will  schedule the exam along with uploading the questions. Student can access the exam . Each student will get randomly shuffled set of questions which he/she has to solve in allotted time. This software will have “Online Remote Proctoring” which will monitor students throughout the examination with the help of  webcam (for laptops). Using Machine Learning The proctor will be constructed to detect head  movement of candidate, face, If it detects any kind of vulnerability, the software will give 3 warnings after which the exam will be automatically submitted by software.

ALGORITHMS:

1. Yolo Algorithm:

YOLO is an algorithm that makes use of neural networks to deliver a period of time face detection. This set of rules is well known because of its speed and accuracy. It has been applied in numerous programs to take a look at traffic signals, people, parking meters, and animals. YOLO rule employs convolutional neural networks (CNN) to take a look at items in a period of time. due to the fact the name suggests, the rule desires totally one forward propagation via a neural community to take a look at items. preceding detection structures repurpose classifiers or localizers to carry out detection. They follow the model to an image at a couple of places and scales. High scoring regions of the image are thought of as detections. We use a very absolutely extraordinary approach. we follow one neural community to the whole photo. This community divides the image into regions and predicts bounding boxes and opportunities for each region. These bounding bins are weighted by the predicted opportunities. Our version has many advantages over classifier-primarily based total structures. it is at the total image at check time thus its predictions are sensible through global context inside the image. It additionally makes predictions with one network analysis in comparison to structures like R-CNN that want thousands for one image. This makes it very speedy, more than 1000x faster than R-CNN and 100x faster than fast R-CNN. See our paper for a whole lot of information about the whole system.

METHODOLOGY OF MULTIPLE LOGIN:-

Define an empty array Time, User, affUser, EvCont, Comp, EvName, Tell Me, Origin

1. and IPaddress For each row with inside the Log document fetch the IPaddress as ‘ip address’ and carry out the following:
2. if ‘ip address’ isn’t always already inserted in IPaddress, then insert it and carry out the following:
3. Insert Date and Time to Time, name to User Full Name to person, affected person to all User, Event Context to EvCont, Component to Comp, Event Name to EvName, Tell approximately were self to Tell Me and origin to origin.
4. Create an output document “output.csv” For each detail in the array ‘consumer’ carry out.
5. Check if there’s replica access for ‘consumer’ is found, if so, then upload all of the respective array entries to the document “output.csv”.
6. The info of the consumer who attempted for more than one logins may be retrieved from the document “output.csv”

OBJECTIVES: –

1. User have to register and login.
2. After registration system start recognizing face for verification.
3. After preparation user start giving exam.
4. While taking exam proctoring system initializing camera and start detecting face and also detecting objects.
5. System will detect if student detect drowsiness while giving exam then system capture the images.

HARDWARE AND SOFTWARE REQUIREMENTS:-

HARDWARE:-

• Processor: Intel Core i3 or more.
• RAM: 4GB or more.
• Hard disk: 250 GB or more.

SOFTWARE:-

• Windows Operating System.
• Java
• R (3.4.1)
• R Studio